Hello Richard,
Odd order distortion is not reduced by making a circuit complementary, balanced or bridged. These techniques work with even order harmonic distortion cancellation.
Regards
Arthur
I like minimalist too 😎, but I don't like NP, sorry.
My target is aural pleasure not minimalist. Less components mean clever engineering.
I actually don't object to the occasional discussion of VFA on this thread, especially when there are those who would like the occasional comparison of CFA to VFA to keep things in perspective. I also think that if a designer is challenged to back up what he says on this thread, then this is the place that the answer should be placed. This, of course, does not preclude Waly from opening another thread dedicated to his new design.
Cheers,
Bob
Richard,
I'm probably going to display my ignorance here, but we don't all know that high frequencies travel mainly on the side edges of a thin rectangular trace, including me.
I have designed PCBs for signals up to 20 Gb/s in my day job on 20-layer Megtron 6 PWBs and I don't have reason to believe that HF travels mainly on the edges. If I'm wrong, I guess I did not do a good job. We have to worry about all kinds of effects, including via stubs and enhanced skin effect losses due to less than perfect copper surface roughness. All HF traces are rounded, not even 45 degree turns are allowed. The path length of the two wires of a balanced line must be managed to be the same within a few mils often (think of doing this when some turns are required or when you have to get in and out of a ball grid array with balls a mil or less apart).
You may know that the best stuff to use these days is Panasonic Megtron 6 (overall, better than Rogers, which we used to use on critical layers). Megtron 6 (and probably some of the Rogers) includes ultra-polished copper surfaces to avoid enhancement of skin effect. At the micron level we are talking about, if the signal has to travel on the surface going up and down "mountains" the effective length of the skin effect path is substantially increased. Bear in mind that the skin depth at 10GHz is way small.
If the HF did not travel on the surface, there would be no reason to do this ultra-smooth surface. In fact, it creates a PIA in that the layer-layer adhesion is reduced with the ultra-polished surface.
Finally, because the edges of the trace are created by an etch, they are usually much rougher than the surface, so their effective skin effect traveling distance is longer.
So please tell me where I am wrong here. I do not consider myself a microwave expert and am always interested in learning more.
Cheers,
Bob
yes... like push-pull. So something else. is there a topology or technique which just cancels odds? A combo?
Thx-RNMarsh
It is not that you are wrong.... rather you missed the importance of the details I mentioned. Such as thickness. If the pcb copper (or plating) is thick 'enough' you wont see the effect. [Though you would if the power levels were increased and moved to higher freq. I discussed this with HP to make a more powerful addition to the test instrument and they were interested and gave some insider support info to me but we never did it]
Rough surface and edges are also a part of the known effects on UHF and contributes to impulse or rise time smearing/group delay. So, yes at high enough freqs this must be improved as well. Note that the effect I am measuring was not at those freqs ---> 100Hz to 100KHz. Where the effect is much weaker and/or can be dominated by other parameters.
The plating on parts or as a conductor - for audio - needs to be at least 1-2mil for audio so that the copper is thick enough to minimize the edge effect. In mass produced audio it is well under 1mil. And, SOTA attempts by DIYers need to be aware of this, IMO. But not if they 'like' distortion etc.... this info isnt for them.
Thx-RNMarsh
Last edited:
Here is a very old thread of mine:
http://www.diyaudio.com/forums/head...h-cascode-headphone-amp-jlh-output-stage.html
Of course it's class A, but nonetheless the input stage is designed so it cancels 3rd harmonics. I suppose I'll have to revisit it.
http://www.diyaudio.com/forums/head...h-cascode-headphone-amp-jlh-output-stage.html
Of course it's class A, but nonetheless the input stage is designed so it cancels 3rd harmonics. I suppose I'll have to revisit it.
Pavel,
Not knowing a lot about the harmonic mathematics here, you saying that you can not have an 8th in isolation, am I correct in interpreting that this would hold for any number that is not a prime number? So this would also hold for a 9th, 3sqrd. not being able to be isolated and that something like numbers that are prime numbers are real values that can be isolated?
Sorry for the way I am saying this, I would have to think about how to say this properly. To long since a math class!
Not knowing a lot about the harmonic mathematics here, you saying that you can not have an 8th in isolation, am I correct in interpreting that this would hold for any number that is not a prime number? So this would also hold for a 9th, 3sqrd. not being able to be isolated and that something like numbers that are prime numbers are real values that can be isolated?
Sorry for the way I am saying this, I would have to think about how to say this properly. To long since a math class!
You may have only 2nd and 3rd pure, in isolation, generated by circuit non-linearity. Higher harmonics appear only in accompaniment for 2nd and 3rd, in a physically real circuit.
yes two waves or a harmonic tone generated as the second tone. BUt, you saw the Audio Precision measuring 8th. Its there when ferrous metals (nonlinear) are part of the circuit.
So instead of denial, understanding and explaining it is needed. Or just dont use it nor have signals in close proxy for your SOTA attempts.
THx-RNMarsh
Last edited:
There was a 2nd and 3rd as well. It was not only 8th in isolation. 8th in isolation is impossible, unless you add it. Impossible as a single product of a circuit non-linearity. I agree, understanding is needed.
?? Who showed 8th in isolation? I didnt.
The 8th is there ONLY when a ferrous metal is introduced in series with the test I/O loop. Similar affect might occur with music frequencies present.
On with the show.
THx-RNMarsh
The 8th is there ONLY when a ferrous metal is introduced in series with the test I/O loop. Similar affect might occur with music frequencies present.
On with the show.
THx-RNMarsh
Last edited:
I am also fine with it as long as we don't sink into another anti CFA diatribe
I think a lot more work is needed before a definitive cause and effect is determined, in any case these artifacts don't excite me as having any bearing on the sound.
It's not anti, it's an ontological debate (call it diatribe, if it makes you feel better).
This is a variant of the JC ubiquitous circuit topology. Is this a CFA or a VFA? Perhaps Mr. Curl would like to jump in and share his opinion?
I have not attempted to explain root cause. This is an affect that occures with ferrous/nonlinear Z metals. One problem I have is scaling these tests up to levels we actually use... many volts and amps. I complained to HP about this.... its like trying to find characteristics of power devices using equipment meant for small signal level devices only.
One can only extrapolate that it would be more and maybe different at higher voltage and currents. In fact, this is the case, but that work comes from the physics world. First exposed to me in research of magnetic field compression R&D. That work has been done and is well known. This is where the edge effect for example came from to me.
Its already known and done. I only wanted to know if I could measure anything that I learned about in the physics research that I was exposed to. And, then do the tests at more typical levels to see if it applies to power amp levels. Then see if the levels are above -100dB to become significant.
BTW- This is the first time and place I ever said anything about this edge delay experiment and how I figured how to do it. I was just doing it for my own intellectual curiosity and knowledge. Approx 15 years ago.
Thx-RNMarsh
One can only extrapolate that it would be more and maybe different at higher voltage and currents. In fact, this is the case, but that work comes from the physics world. First exposed to me in research of magnetic field compression R&D. That work has been done and is well known. This is where the edge effect for example came from to me.
Its already known and done. I only wanted to know if I could measure anything that I learned about in the physics research that I was exposed to. And, then do the tests at more typical levels to see if it applies to power amp levels. Then see if the levels are above -100dB to become significant.
BTW- This is the first time and place I ever said anything about this edge delay experiment and how I figured how to do it. I was just doing it for my own intellectual curiosity and knowledge. Approx 15 years ago.
Thx-RNMarsh
Last edited:
Richard,
You have to remember that with Eastern music and a Luxman radio perhaps none of that is important? Harmonics may be one of the desirable attributes of that music?
You have to remember that with Eastern music and a Luxman radio perhaps none of that is important? Harmonics may be one of the desirable attributes of that music?